June 3, 2020

Explorative Analysis

Objective

Policymakers in different Countries have introduced different political action to contrast the COVID19 contagion.

  1. What are the different containment efforts and is there a strategies resemblance across countries?

  2. What is the effect of these policies on the contagion from a global perspective?

  3. Has the same action lead to different results in the case of different regions of Italy?

Data

  • COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University for contagion data,

  • Oxford COVID-19 Government Response Tracker (OxCGRT) for policies tracking.

Containment strategies and resembling patterns

  • Dimension Reduction via Polychoric PCA for \(11\) ordinal variables (from \(0\) to \(2\) or \(0\) to \(3\)) indicating the stringency level of policies such as

    • School, workplace and transport closing and event cancellation;

    • Gathering, stay-home and internal/international movement restrictions;

    • Information, Testing and Contact Tracing campaigns.

  • Functional Data Co-Clustering of the countries aligned to the first contagion (from the 10th day before contagion).

Containment strategies

Containment strategies

Restriction-based policies on one hand, Tracing and Testing policies on the other hand.

Resembling patterns

Resembling patterns

Worldwide Analysis

Motivation

Model

GENERALIZED POISSON MIXED MODEL for Overdispersed Count Data
  • We analyze the number of Active person, i.e., Confirmed - Deaths - Recovered \(\rightarrow\) Count Dependent Variable (Generalized Poisson Model);

  • The data are observed for each country nested within clusters during \(103\) days \(\rightarrow\) Mixed Model.

  • Confounders from the World Bank Open Data and Oxford Data:
    • FIXED:
      • Population density;
    • LONGITUDINAL:
      • Economic: Income Support and Debt/contract relief for households;
      • Health: Emergency Investment in healthcare and Investment in vaccines.

Which policies acted better?

Who acted promptly?

Who acted promptly?

Take home message

  • Lockdown policies work! respect to impose no measure;

  • Strong Testing and Tracing policies lead to discovering more infected people (luckily!)

  • Korea and Singapore are the best countries that acted properly;

  • Sweden, Germany, Portugal, and Greece better than the other UE countries;

  • The USA, and Canada better than the other UE countries except for Sweden and Germany.

Italy Case

Introduction

Background
  • Italian regions, ethernal divide

  • Lockdown almost simultaneous, excepted the Red Zone

  • First cases in Lombardia and Lazio hubs

Problems
  • Policies have no variability between regions

  • Baseline control: some regions start from worse situations

  • Cannot estimates some effects as for the nations case

  • To our defense, integration between databases came lately

  • Instrumental variables, more correct but tricky approach

Approach

  • Phase “1” versus Phase “0” comparison

  • Auto-regression: modelling active cases given past numbers
    • Related but not quite to the R0 index
  • Random effects for region and date, standard panel approach

  • Assuming policies effects seen ~14 days later

  • Controlling for testing frequency

Speed of contagion

ETV, Estimated Time to Victory

Joint view

A map of criticality

Anything weird?

Conclusions